Loading... Please wait...





    XILS 3 :
    "The sound of XILS 3 is amazing, finally we can have an EMS in the studio and fulfil all our BBC Radiophonic Workshop dreams"
    Evil Nine

One of its features, never heard on other vintage analog synthesizers, was the continuously modulated waveform, sort of a modulated waveform morphing. Recreating this feature in the digital domain, without aliasing was a very big challenge.




Why was the polyKobol not done before in software? The fact that only 30 were ever made and they came out at the same time as the DX7 might have had some influence on how obscure this synth is?


Yes maybe, but I think that the specific oscillators this machine brings was one of the main reason why it never been done as a virtual instrument .


knob_waveform.png  Doing an oscillator in the digital domain implies finding a workaround to avoid aliasing. This aliasing appears when the frequency of the oscillator is not a multiple of the sample frequency. Various methods can be found, included tables or Band Limited Impulse Train. But these two methods can't achieve good results when PWM must be done or when waveforms are not standard (like pulse, saw or triangle). Furthermore when hard synchronization between oscillators is needed, these methods fail.


Since most of actual Virtual Instrument's oscillators are using one or the other method, the polyKobol, despite its wonderful sounds possibilities was never reproduced in a software product.

So the challenge was very big : How to generate such of waveform, providing PWM ?

First of all, I need to know how this synthesizer works, how it sounds what was its character, but with a few units working around the world, it wasn't obvious to find information.

Hopefully I discovered a web-site where the polyKobol and other RSF synthesizers were described : 


So I contacted Olivier Grall who launched this web-site and who kindly, answered to my questions. As an RSF synthesizer specialist who own a polyKobol II (and other few RSF synthesizers) he gave me much of the informations needed, as well as some very accurate samples. Thanks to this, I was able to begin my recreation work.

You mean DSP programming ...

Yes, but the beginning was very hard : all the methods I tried didn't work correctly. No way to use standard waveforms and to mix them, no way to use look-up table table, this would reduce drastically the hard synchronization and modulation possibilities.

So as I did with the XILS 3, I take the problem from the very beginning, I mean from the theoretical BLIT (Band Limited Impulsion Train) method.

Developing this synthesizer was a labor of love....I knew I could do it...and I knew why others would never try. It took a lot of work and focus to get the sound. I am glad I am not easily discouraged

So I stretched the algorithm and turned it in all senses and after a lot of work, I achieved my goal : the oscillators can morph smoothly from triangle to pulse, passing by saw with a clear and deep sound, without aliasing.

And the hard synchronization ?


This process presented a whole new set of challenges. Stretching the algorithm again, turn it in all senses again, but finally, magically, it was done. The first hard synchronization between two oscillators, generating PWM morphed sound was born.


polyKB's oscillators hard synced